This proposal seeks support for a series of studies designed to help us to understand, evaluate, and eventually emulate the strategies and techniques that exceptional human tutors use to motivate and to instruct children. Its aim is to develop a model of the procedures that expert tutors use not only to increase children's academic accomplishments, but also to enhance children's social adjustment to school by enhancing their academic self-esteem, their perceptions of self-efficacy, their feelings of personal control, their Intrinsic interest In learning, and their willingness to seek out and persist at difficult tasks In the future. In particular, the project involves the collection of a library of videotaped protocols of demonstrably outstanding tutors as they work with students on topics in elementary mathematics. Following each tutoring session, these tutors will also be asked to watch the tape of that session and to describe the reasons underlying their decisions on how, why, where, and when to intervene. Learners, likewise, will watch tapes of their sessions and describe their thoughts and feelings and their Impressions of the tutor. These data sets will be contrasted with comparable tutoring protocols and stimulated-recall interviews with other experienced, but less effective, tutors and with novice tutors. Moreover, all tutors will be observed working with two samples of students, selected to represent groups educationally 'at risk' and 'not at risk.' The analysis will focus particularly on the Interplay of cognitive and motivational goals and the intellectual and the affective and motivational consequences for children of working with outstanding tutors. The general goal will be to construct an integrated model of both the cognitive and the affective goals and strategies of expert tutors in early mathematics, such that one can derive from a hierarchy of motivational and instructional goals the central considerations that govern the control, the timing, the content, and the form of tutors' statements to both 'at risk' and 'not at risk' students. The development of such a model, in turn, should eventually help us to develop better methods for teaching novices and volunteers to become effective tutors, to derive more effective techniques for small-group instruction, and to design more powerful computer-based tutoring systems that will take into account affective and motivational, as well as cognitive and informational, aspects of the tutorial process.